Capstone #1¶
File Descriptions¶
- DATA_ELEMENT_DESCRIPTION.csv defines each data element and indicates where its description is found in Data Sources, Definitions, and Notes.
- DEFINED_DATA_VALUE.csv defines the meaning of specific values (such as missing or suppressed data).
- DEMOGRAPHICS.csv identifies the data elements and values in the Demographics indicator domain.
- HEALTHY_PEOPLE_2010.csv identifies the Healthy People 2010 Targets and the U.S. Percentages or Rates.
- LEADING_CAUSES_OF_DEATH.csv identifies the data elements and values in the Leading Causes of Death indicator domain.
- MEASURES_OF_BIRTH_AND_DEATH.csv identifies the data elements and values in the Measures of Birth and Death indicator domain.
- PREVENTIVE_SERVICES_USE.csv identifies the data elements and values in the Preventive Services indicator domain.
- RELATIVE_HEALTH_IMPORTANCE.csv identifies the data elements and values in the Relative Health Importance indicator domain.
- RISK_FACTORS_AND_ACCESS_TO_CARE.csv identifies the data elements and values in the Risk Factors and Access to Care indicator domain.
- SUMMARY_MEASURES_OF_HEALTH.csv identifies the data elements and values in the Summary Measures of Health indicator domain.
- VULNERABLE_POPS_AND_ENV_HEALTH.csv identifies the data elements and values in the Vulnerable Populations and Environmental Health indicator domain.
Data source can be found here
In [1]:
import glob
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib as mpl
import matplotlib.cm as cm
from matplotlib.pyplot import figaspect
from mpl_toolkits.axes_grid1 import make_axes_locatable
import geopandas as gpd
import seaborn as sns
from scipy.stats.stats import pearsonr
mpl.rcParams.update({'font.size': 14})
In [2]:
#Reading all 11 csv files
files = sorted(glob.glob('../CHSI_dataset/*.csv'))
File Initialization¶
In [3]:
DATA_ELEMENT_DESCRIPTION = pd.read_csv(files[0])
DEFINED_DATA_VALUE = pd.read_csv(files[1])
DEMOGRAPHICS = pd.read_csv(files[2])
HEALTHY_PEOPLE_2010 = pd.read_csv(files[3])
LEADING_CAUSES_OF_DEATH = pd.read_csv(files[4])
MEASURES_OF_BIRTH_AND_DEATH = pd.read_csv(files[5])
PREVENTIVE_SERVICES_USE = pd.read_csv(files[6])
RELATIVE_HEALTH_IMPORTANCE = pd.read_csv(files[7])
RISK_FACTORS_AND_ACCESS_TO_CARE = pd.read_csv(files[8])
SUMMARY_MEASURES_OF_HEALTH = pd.read_csv(files[9])
VULNERABLE_POPS_AND_ENV_HEALTH = pd.read_csv(files[10])
In [4]:
#Auxiliary files
DATA_ELEMENT_DESCRIPTION
DEFINED_DATA_VALUE
HEALTHY_PEOPLE_2010
Out[4]:
In [5]:
LEADING_CAUSES_OF_DEATH.head()
Out[5]:
In [6]:
#Outlier column name
DATA_ELEMENT_DESCRIPTION.update(
DATA_ELEMENT_DESCRIPTION.replace('C_Ot_homicide','C_Ot_Homicide'))
LEADING_CAUSES_OF_DEATH.rename(columns={'C_Ot_homicide':'C_Ot_Homicide'}, inplace=True)
#Functions to describe acronyms
def description(param):
if param in list(DATA_ELEMENT_DESCRIPTION.COLUMN_NAME):
description_value = DATA_ELEMENT_DESCRIPTION[DATA_ELEMENT_DESCRIPTION.
COLUMN_NAME==param].DESCRIPTION.tolist()[0]
else:
description_value = ''
return description_value
def short_description(param):
list_of_removing = ['County data,', 'death measures,', 'birth measures,', ',']
result = description(param)
for i in list_of_removing:
if result.__contains__(i):
result = result.replace(i,'')
return result
def short_race_description(param):
list_of_removing = ['Black', 'Hispanic','White', 'other',
'County data,', 'death measures,', 'birth measures,', ',']
result = description(param)
for i in list_of_removing:
if result.__contains__(i):
result = result.replace(i,'')
return result
In [7]:
#Common columns in all worksheets
county_info = ['State_FIPS_Code', 'County_FIPS_Code', 'CHSI_County_Name',
'CHSI_State_Name', 'CHSI_State_Abbr', 'Strata_ID_Number']
In [8]:
#Joining files
worksheets = [DEMOGRAPHICS,
LEADING_CAUSES_OF_DEATH,
MEASURES_OF_BIRTH_AND_DEATH,
PREVENTIVE_SERVICES_USE,
RELATIVE_HEALTH_IMPORTANCE,
RISK_FACTORS_AND_ACCESS_TO_CARE,
SUMMARY_MEASURES_OF_HEALTH,
VULNERABLE_POPS_AND_ENV_HEALTH]
In [9]:
#Replacing not available/reported data by zeros
nan_values = [-1111, -1111.1, -1, -9999, -2222, -2222.2, -2]
for worksheet in worksheets:
nda =[]
for i,nan_value in enumerate(nan_values):
nda.append(worksheet.loc[:,:]!=nan_value)
worksheet.update(worksheet.where(nda[i]).fillna(0))
In [10]:
#Time dependent data should be identified and normalized
time_dependent_worksheets = [LEADING_CAUSES_OF_DEATH,
MEASURES_OF_BIRTH_AND_DEATH,
PREVENTIVE_SERVICES_USE,
VULNERABLE_POPS_AND_ENV_HEALTH]
time_spans = ['1994-2003', '1999-2003', '2001-2003']
time_span_convert = [7,4,2]
for worksheet in time_dependent_worksheets:
for i in list(worksheet.columns):
if i.__contains__('Time_Span'):
worksheet[i].replace(time_spans, time_span_convert, inplace=True)
worksheet.rename(columns={i: 'Time'}, inplace=True)
for worksheet in time_dependent_worksheets:
for i in list(worksheet.columns):
if worksheet.columns.get_loc(i) < worksheet.columns.get_loc('Time') and i not in county_info:
worksheet[i]=worksheet[i]/worksheet['Time']
In [11]:
#Merging all clean data files
df = worksheets[0]
for name in worksheets[1:]:
df = df.merge(name, on=county_info.append('Time'), how='outer', sort=True).fillna(0)
In [12]:
df.head()
Out[12]:
Redundant columns¶
In [13]:
observation = list(df.columns)
description_list = [description(i) for i in observation]
def observ_keyword(variable):
results = []
for i, j in enumerate(description_list):
if variable in str(j):
results.append(observation[i])
return results
In [14]:
filter_percentile = list(description(i) for i in list(observ_keyword('percentile')))
filter_CI = list(description(i) for i in list(observ_keyword('Confidence interval')))
filter_fav = list(description(i) for i in list(observ_keyword('Favorable indicator')))
filters = list(observ_keyword('Favorable indicator')
+ observ_keyword('Confidence interval')
+ observ_keyword('percentile'))
df = df.drop(columns=filters)
In [15]:
#Neutral columns
for i in list(df.columns):
if len(np.unique(df[i]))==1 and i!='D_Ot_HIV':
df=df.drop(columns=i)
df.head()
Out[15]:
In [16]:
df.describe().transpose()
Out[16]:
Demographic Map¶
To visualize the health indices throughout the country, I merge the map shapefile and data to us_merge dataframes. Because Alaska and Hawaii are further away and out of proportion I merged them separately. US_plot function creates the US map with any numeric variables in the data --default color is blue.
In [17]:
us_map = gpd.read_file('../US-map/UScounties.shp')
us_map_land = us_map[ (us_map['STATE_NAME']!='Alaska')
&(us_map['STATE_NAME']!='Hawaii')]
us_map_Alaska = us_map[us_map['STATE_NAME']=='Alaska']
us_map_Hawaii = us_map[us_map['STATE_NAME']=='Hawaii']
#Merging map and data:
df_land = us_map_land.set_index('NAME').join(df.set_index('CHSI_County_Name'))
df_Alaska = us_map_Alaska.set_index('NAME').join(df.set_index('CHSI_County_Name'))
df_Hawaii = us_map_Hawaii.set_index('NAME').join(df.set_index('CHSI_County_Name'))
In [18]:
def min_max_col(variable):
minimum = min(df_land[variable].min(),
df_Alaska[variable].min(), df_Hawaii[variable].min())
maximum = max(df_land[variable].max(),
df_Alaska[variable].max(), df_Hawaii[variable].max())
return minimum, maximum
def US_plot(variable, color='Blues'):
h, w = figaspect(1.)
mn, mx = min_max_col(variable)
fig, ax1 = plt.subplots(1, figsize=(w*10, h*10))
ax1.axis('off')
ax1.set_title(description(variable), fontsize=40)
divider = make_axes_locatable(ax1)
cax = divider.append_axes('right', size='2%', pad=0.1)
cax.tick_params(labelsize=25)
df_land.plot(column=variable, cmap=color,
linewidth=0.8, ax=ax1,
edgecolor='0.8', vmin=mn ,vmax=mx,
legend=True, cax=cax)
left, bottom, width, height = [0.15, 0.22, 0.24, 0.3]
ax2 = fig.add_axes([left, bottom, width, height])
ax2.axis('off')
ax2.set_title('Alaska', fontsize=25)
df_Alaska.plot(column=variable, cmap=color, vmin=mn ,vmax=mx,
linewidth=0.8, ax=ax2, edgecolor='0.6')
left, bottom, width, height = [0.35, 0.3, 0.2, 0.05]
ax3 = fig.add_axes([left, bottom, width, height])
ax3.axis('off')
ax3.set_title('Hawaii', fontsize=25)
df_Hawaii.plot(column =variable, cmap=color, vmin=mn ,vmax=mx,
linewidth=0.8, ax=ax3, edgecolor='0.6')
In [19]:
US_plot('Suicide', 'Greys')
In [20]:
US_plot('CHD', 'Reds')
Measures of Birth and Death¶
In [21]:
Birth_Death_list = list(np.intersect1d(df.columns, MEASURES_OF_BIRTH_AND_DEATH.columns))
Birth = []
Death = []
for i in Birth_Death_list:
if i not in county_info:
if description(i).__contains__('birth'):
Birth.append(i)
else:
Death.append(i)
In [50]:
Death_Mean = {'Cause_of_Death':Death, 'Mean_Death':[np.mean(df[i]) for i in Death]}
Birth_Mean = {'Birth':Birth, 'Mean_Birth':[np.mean(df[i]) for i in Birth]}
df_death_mean = pd.DataFrame(Death_Mean)
df_death_mean = df_death_mean.drop(df_death_mean[df_death_mean['Cause_of_Death'] =='Total_Deaths'].index, axis=0)
df_death_mean = df_death_mean.sort_values(by=['Mean_Death'], ascending=False)
df_birth_mean = pd.DataFrame(Birth_Mean)
df_birth_mean = df_birth_mean.drop(df_birth_mean[df_birth_mean['Birth'] =='Total_Births'].index, axis=0)
df_birth_mean = df_birth_mean.sort_values(by=['Mean_Birth'], ascending=False)
In [51]:
fig, ax = plt.subplots(1, figsize=(5, 5))
labels=list(short_description(i) for i in df_death_mean.Cause_of_Death)
plt.barh(labels, df_death_mean.Mean_Death, color='brown', height=0.9)
plt.xlabel('Mean of death in each county', fontsize=12)
plt.title('Cause of death rates', y=0.845, fontsize=14,
bbox=dict(facecolor='none', edgecolor='k', boxstyle='round,pad=0.5'))
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.show()
In [53]:
fig, ax = plt.subplots(1, figsize=(4, 4))
labels=list(short_description(i) for i in df_birth_mean.Birth)
plt.barh(labels, df_birth_mean.Mean_Birth, color='b', height=0.9)
plt.xlabel('Mean of birth on each county', fontsize=12)
plt.title('Birth conditions', y=0.872, fontsize=14,
bbox=dict(facecolor='none', edgecolor='k', boxstyle='round,pad=0.5'))
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.show()
In [25]:
Death_race_list = list(np.intersect1d(df.columns, LEADING_CAUSES_OF_DEATH.columns))
Black = []
Hispanic = []
White = []
other = []
for i in Death_race_list:
if i not in county_info:
if description(i).__contains__('Black'):
Black.append(i)
elif description(i).__contains__('Hispanic'):
Hispanic.append(i)
elif description(i).__contains__('White'):
White.append(i)
else:
other.append(i)
In [26]:
Black_up_24=[]
Black_25=[]
Hispanic_up_24=[]
Hispanic_25=[]
White_up_24=[]
White_25=[]
other_up_24=[]
other_25=[]
for i in Black:
if description(i).__contains__('25-44') or description(i).__contains__('45-64') or description(i).__contains__('65+'):
Black_25.append(i)
else:
Black_up_24.append(i)
for i in Hispanic:
if description(i).__contains__('25-44') or description(i).__contains__('45-64') or description(i).__contains__('65+'):
Hispanic_25.append(i)
else:
Hispanic_up_24.append(i)
for i in White:
if description(i).__contains__('25-44') or description(i).__contains__('45-64') or description(i).__contains__('65+'):
White_25.append(i)
else:
White_up_24.append(i)
for i in other:
if description(i).__contains__('25-44') or description(i).__contains__('45-64') or description(i).__contains__('65+'):
other_25.append(i)
else:
other_up_24.append(i)
In [27]:
fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2,2, figsize=(25, 25), sharey=True)
fig.subplots_adjust(hspace=-0.1)
cs=cm.Pastel1(np.arange(100))
mpl.rcParams['font.size'] = 15
fig.suptitle('Cause of Death, age under 25: Different Races', y=0.95, fontsize=30)
ax1.set_title('Black', fontsize=25)
labels=list(short_race_description(i) for i in Black_up_24)
ax1.pie(list(np.mean(df[i]) for i in Black_up_24),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
ax2.set_title('Hispanic', fontsize=25)
labels=list(short_race_description(i) for i in Hispanic_up_24)
ax2.pie(list(np.mean(df[i]) for i in Hispanic_up_24),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
ax3.set_title('White', fontsize=25)
labels=list(short_race_description(i) for i in White_up_24)
ax3.pie(list(np.mean(df[i]) for i in White_up_24),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
ax4.set_title('Other', fontsize=25)
labels=list(short_race_description(i) for i in other_up_24)
ax4.pie(list(np.mean(df[i])*2 for i in other_up_24),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
plt.show()
In [45]:
fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2,2, figsize=(25, 25), sharey=True)
fig.subplots_adjust(hspace=-0.1)
cs=cm.Set3(np.arange(100))
mpl.rcParams['font.size'] = 15
fig.suptitle('Cause of Death, age 25+: Different Races', y=0.95, fontsize=30)
ax1.set_title('Black', fontsize=25)
labels=list(short_race_description(i) for i in Black_25)
ax1.pie(list(np.mean(df[i]) for i in Black_25),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
ax2.set_title('Hispanic', fontsize=25)
labels=list(short_race_description(i) for i in Hispanic_25)
ax2.pie(list(np.mean(df[i]) for i in Hispanic_25),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
ax3.set_title('White', fontsize=25)
labels=list(short_race_description(i) for i in White_25)
ax3.pie(list(np.mean(df[i]) for i in White_25),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
ax4.set_title('Other', fontsize=25)
labels=list(short_race_description(i) for i in other_25)
ax4.pie(list(np.mean(df[i]) for i in other_25),
labels=labels, colors=cs, autopct='%1.1f%%', radius=0.8)
plt.show()
In [30]:
df.Mammogram
US_plot('Mammogram')
In [31]:
US_plot('Brst_Cancer')
Correlations between Variables¶
The following table shows the Pearson correlation between variables:
In [54]:
df.corr(method ='pearson')
Out[54]:
Well, this is too much information about our big dataframe! To know about the important correlated variables we need more cleaning;
I'll make another small dataframe with 3 columns: column one and two--> correlated pairs and column 3 the magnitude of their correlations. To make it easier to see the highly correlated parameters I sort them. I also defined a threshold=0.7 to cut low-correlated ones.
In [33]:
not_helping_columns = ['Sulfur_Dioxide_Ind',
'Nitrogen_Dioxide_Ind', 'Lead_Ind', 'Carbon_Monoxide_Ind', 'Particulate_Matter_Ind', 'Time']
attrs = df.drop(columns=not_helping_columns).corr()
threshold = 0.6
high_corrs = (attrs[abs(attrs) > threshold][attrs != 1.0]).unstack().dropna().to_dict()
df_corrs = pd.DataFrame(list((sorted(key)[0], sorted(key)[1], high_corrs [key]) for key in high_corrs),
columns=['attribute_1','attribute_2', 'correlation'])
df_corrs = df_corrs.iloc[abs(df_corrs['correlation']).argsort()[::-1]].drop_duplicates().reset_index(drop=True)
df_corrs.head(20)
Out[33]:
Alright! It looks better!
In [55]:
# If we want to make a df with tuples of correlated pairs:
# df_corrs = pd.DataFrame(list(set([(tuple(sorted(key)), high_corrs [key])
# for key in high_corrs])), columns=['attribute_pair', 'correlation'])
In [34]:
#Here we plot the top 6 most correlated parameters:
colors = ['b', 'g', 'r', 'c', 'm', 'y']
markers = ['H', 'v', 's', 'P', '*', 'D']
fig, axs = plt.subplots(3,2, figsize=(12,15), facecolor='w', edgecolor='k')
fig.subplots_adjust(hspace=0.4, wspace=0.5)
fig.suptitle('Plots of highly-correlated parameters', y=0.97, fontsize=30)
axs = axs.ravel()
for i in range(6):
axs[i].scatter(df[df_corrs.loc[i+1,'attribute_1']], df[df_corrs.loc[i+1,'attribute_2']],
marker=markers[i], color=colors[i])
axs[i].set_xlabel(short_description(df_corrs.loc[i+1,'attribute_1']), fontsize=14)
axs[i].set_ylabel(short_description(df_corrs.loc[i+1,'attribute_2']), fontsize=14)
axs[i].tick_params(axis='x', labelsize=10)
axs[i].tick_params(axis='y', labelsize=10)
In [35]:
for i in Death:
print(df_corrs[df_corrs['attribute_1']==i].head(10))
In [36]:
#Now the most correlated parameters with heart diseas:
num_plots = 16
colors = cm.tab20(np.linspace(0, 1, num_plots))
fig, axs = plt.subplots(int(num_plots/2),2, figsize=(12,40), facecolor='w', edgecolor='k')
fig.subplots_adjust(hspace=0.4, wspace=0.5)
fig.suptitle('Plots of most correlated parameters with heart disease', y=0.92, fontsize=30)
axs = axs.ravel()
corr_params = list(df_corrs[df_corrs['attribute_1']=='CHD']['attribute_2'])
for i in range(num_plots):
axs[i].scatter(df['CHD'], df[corr_params[i]],
marker='*', color=colors[i])
axs[i].set_xlabel(short_description('CHD'), fontsize=14)
axs[i].set_ylabel(short_description(corr_params[i]), fontsize=14)
axs[i].tick_params(axis='x', labelsize=10)
axs[i].tick_params(axis='y', labelsize=10)
In [37]:
#Now the most correlated parameters with breast cancer:
num_plots = 10
colors = cm.tab20(np.linspace(0, 1, num_plots))
fig, axs = plt.subplots(int(num_plots/2),2, figsize=(12,25), facecolor='w', edgecolor='k')
fig.subplots_adjust(hspace=0.4, wspace=0.5)
fig.suptitle('Plots of most correlated parameters with breast cancer', y=0.92, fontsize=30)
axs = axs.ravel()
corr_params = list(df_corrs[df_corrs['attribute_1']=='Brst_Cancer']['attribute_2'])
for i in range(num_plots):
axs[i].scatter(df['CHD'], df[corr_params[i]],
marker='*', color=colors[i])
axs[i].set_xlabel(short_description('Brst_Cancer'), fontsize=14)
axs[i].set_ylabel(short_description(corr_params[i]), fontsize=14)
axs[i].tick_params(axis='x', labelsize=10)
axs[i].tick_params(axis='y', labelsize=10)
Okk! Correlated parameters are great and can be discussed in statistical inference section. Now how about looking at the data from States point of view:
In [38]:
#A new dataframe based on States:
df_state = df.groupby('CHSI_State_Abbr').sum().reset_index()
df_state.head()
Out[38]:
In [39]:
sns.set(style='whitegrid')
fig, ax = plt.subplots(1, figsize=(20, 10))
plt.bar(df_state['CHSI_State_Abbr'], df_state['Homicide'], color='b')
plt.xlim(-1, len(df_state.index))
plt.ylabel('The sum of homicide in all the counties of each state', fontsize=16)
plt.xticks(fontsize=12)
plt.show()
In [40]:
sns.reset_defaults()
us_state = gpd.read_file('../US-States/states.shp')
us_state_land = us_state[ (us_state['STATE_NAME']!='Alaska')
&(us_state['STATE_NAME']!='Hawaii')]
us_state_Alaska = us_state[us_state['STATE_NAME']=='Alaska']
us_state_Hawaii = us_state[us_state['STATE_NAME']=='Hawaii']
us_state.head()
#Merging map and data:
df_state_land = us_state_land.set_index('STATE_ABBR').join(df_state.set_index('CHSI_State_Abbr'))
df_state_Alaska = us_state_Alaska.set_index('STATE_ABBR').join(df_state.set_index('CHSI_State_Abbr'))
df_state_Hawaii = us_state_Hawaii.set_index('STATE_ABBR').join(df_state.set_index('CHSI_State_Abbr'))
def min_max_col2(variable):
minimum = min(df_state_land[variable].min(),
df_state_Alaska[variable].min(), df_state_Hawaii[variable].min())
maximum = max(df_state_land[variable].max(),
df_state_Alaska[variable].max(), df_state_Hawaii[variable].max())
return minimum, maximum
def US_state_plot(variable, color='Blues'):
h, w = figaspect(1.)
mn, mx = min_max_col2(variable)
fig, ax1 = plt.subplots(1, figsize=(w*10, h*10))
ax1.axis('off')
ax1.set_title(short_description(variable), fontsize=40)
divider = make_axes_locatable(ax1)
cax = divider.append_axes('right', size='2%', pad=0.1)
cax.tick_params(labelsize=25)
df_state_land.plot(column=variable, cmap=color,
linewidth=0.8, ax=ax1,
edgecolor='0.8', vmin=mn ,vmax=mx,
legend=True, cax=cax)
left, bottom, width, height = [0.15, 0.22, 0.24, 0.3]
ax2 = fig.add_axes([left, bottom, width, height])
ax2.axis('off')
ax2.set_title('Alaska', fontsize=25)
df_state_Alaska.plot(column=variable, cmap=color, vmin=mn ,vmax=mx,
linewidth=0.8, ax=ax2, edgecolor='0.6')
left, bottom, width, height = [0.35, 0.3, 0.2, 0.05]
ax3 = fig.add_axes([left, bottom, width, height])
ax3.axis('off')
ax3.set_title('Hawaii', fontsize=25)
df_state_Hawaii.plot(column =variable, cmap=color, vmin=mn ,vmax=mx,
linewidth=0.8, ax=ax3, edgecolor='0.6')
In [41]:
df_state_land['Homicide_per_capita'] = df_state_land['Homicide'].div(df_state_land['Population_Size'])
df_state_Alaska['Homicide_per_capita'] = df_state_Alaska['Homicide'].div(df_state_Alaska['Population_Size'])
df_state_Hawaii['Homicide_per_capita'] = df_state_Hawaii['Homicide'].div(df_state_Hawaii['Population_Size'])
new_item = pd.DataFrame({'COLUMN_NAME':['Homicide_per_capita'], 'DESCRIPTION':['Homicide per capita']})
DATA_ELEMENT_DESCRIPTION = DATA_ELEMENT_DESCRIPTION.append(new_item, sort=False)
US_state_plot('Homicide_per_capita', 'OrRd')
In [42]:
df_state_land['Suicide_per_capita'] = df_state_land['Suicide']/df_state_land['Population_Size']
df_state_Alaska['Suicide_per_capita'] = df_state_Alaska['Suicide']/df_state_Alaska['Population_Size']
df_state_Hawaii['Suicide_per_capita'] = df_state_Hawaii['Suicide']/df_state_Hawaii['Population_Size']
new_item = pd.DataFrame({'COLUMN_NAME':['Suicide_per_capita'], 'DESCRIPTION':['Suicide per capita']})
DATA_ELEMENT_DESCRIPTION = DATA_ELEMENT_DESCRIPTION.append(new_item, sort=False)
US_state_plot('Suicide_per_capita', 'BuPu')
In [93]:
#Relationship between total death, being unmarried and being uninsured:
sns.set(style='whitegrid')
fig, ax = plt.subplots(1, figsize=(6, 5))
cmap = sns.cubehelix_palette(light=1, as_cmap=True)
ax = sns.scatterplot(x='Unemployed', y='Uninsured', size='Total_Deaths', hue='Total_Deaths',
sizes=(20, 200), palette=cmap, data=df_state)
ax.set_xlim(0)
ax.set_ylim(0)
Out[93]:
In [100]:
sns.reset_orig()
f, ax = plt.subplots(figsize=(5, 4))
ax = sns.regplot(x='Unmarried', y='Infant_Mortality', data=df_state)
ax.set(ylim=(0, None))
ax.set(xlim=(0, None))
ax.set_ylabel('Infant Mortality')
Out[100]:
